Diving deeper into Administration and Governance

Objective

After completing this lesson, you will be able to provision Event Broker Services tailored to performance, storage, and region requirements

Sizing and Provisioning Event Broker Services

Before deploying event-driven messaging infrastructure, platform engineers must carefully evaluate system workloads, performance requirements, and deployment regions. This concept topic explores how to select appropriate cloud providers, choose right-sized Service Classes, and automate the lifecycle management of Event Broker Services through the Cloud Console.

Selecting Cloud Providers and Regions

Deploying messaging infrastructure in SAP Integration Suite, advanced event mesh requires careful architectural planning before triggering automated provisioning workflows. Administrators must align the messaging infrastructure with expected application workloads, performance requirements, and regulatory boundaries.

The platform offers full deployment flexibility across major public cloud hyperscalers, such as Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP), as well as Dedicated Regions deployed on-premises or within private cloud data centers. Choosing the optimal cloud region is driven by three primary operational considerations:

Geographic Proximity and Latency

Co-locating Event Broker Services in the same cloud region as primary event producers (such as SAP S/4HANA Cloud instances or core microservices) drastically reduces round-trip network latency, accelerating real-time message processing.

Data Transfer Costs

Minimizing cross-region and cross-cloud traffic lowers egress bandwidth fees, keeping enterprise integration costs predictable.

Regulatory Compliance and Data Sovereignty

Certain legal frameworks require sensitive operational data and message payload spools to remain strictly within specific geographical boundaries.

Customer-Controlled Regions extend Cloud Console management into private data centers to enforce compliance without sacrificing cloud-native administrative workflows.

Sizing and Provisioning

Service Sizing and Service Classes

Service Classes define the hardware allocation, compute capabilities, and capacity boundaries of an Event Broker Service. Administrators select a Service Class based on four core performance metrics:

Concurrent Client Connections

The total number of simultaneous applications, IoT devices, or integration endpoints connected to the broker.

Message Throughput

The maximum volume of incoming and outgoing messages the service must process per second.

Message Spool Storage

Dedicated persistent disk capacity allocated for holding unacknowledged or queued messages during consumer offline periods.

Redundancy Requirements

Non-production testing environments utilize standalone developer tiers to conserve costs, whereas production environments require high-availability enterprise tiers operating across multiple availability zones.

Automated Provisioning and Lifecycle Management

Once service parameters are selected, administrators launch provisioning directly from the Cloud Console. The platform automates underlying infrastructure orchestration, provisions dedicated messaging nodes, applies default security profiles, and assigns distinct connection endpoints for supported protocols including AMQP, MQTT, REST, and Solace Web Messaging. Post-deployment, administrators manage running instances throughout their lifecycle, scaling spool storage dynamically, upgrading Service Classes to handle application growth, applying zero-downtime platform updates, or decommissioning inactive instances to release cloud resources.

Connecting Hybrid Topologies with Dynamic Message Routing

Connecting standalone broker services into a unified, enterprise-wide Event Mesh enables seamless real-time event distribution across multi-cloud and hybrid environments. This concept topic covers the configuration of Mesh Links and the inner workings of Dynamic Message Routing (DMR) for bandwidth-efficient message delivery.

Establishing Mesh Links

An Event Mesh is a configurable network of interconnected Event Broker Services that distributes real-time events seamlessly across hybrid and multi-cloud landscapes. It liberates applications from maintaining static point-to-point network connections by allowing producers to publish events locally while the mesh handles global distribution.

The backbone of an Event Mesh consists of Mesh Links—secure, high-throughput inter-broker communication channels established between nodes operating in different cloud regions or data centers. Administrators configure Mesh Links via guided wizards in the Cloud Console by defining target broker endpoints, transport security parameters (TLS encryption), and mutual authentication credentials. Once established, Mesh Links serve as multi-protocol bridges that transparently carry event traffic between distributed environments.

Mesh Links

Dynamic Message Routing (DMR) Operations

Rather than requiring complex static routing tables or manually maintained routing rules, the event mesh relies on Dynamic Message Routing (DMR). DMR continuously coordinates subscription states across the entire broker network.

Connected brokers exchange active topic subscriptions in real time across Mesh Links. When a publisher posts an event to Broker A, Broker A inspects the global subscription topology. If an application connected to Broker B has an active subscription for that event topic, Broker A routes the message across the Mesh Link to Broker B.

If no connected applications on Broker C subscribe to that topic, the message is never transmitted to Broker C. This intelligent filtering prevents unnecessary network traffic from crossing expensive WAN or cloud provider boundaries, maximizing performance and minimizing bandwidth expenditure.

High Availability, Fault Tolerance, and Event Preservation

Enterprise integration flows require high availability and robust error-handling capabilities to prevent data loss during system outages or application failures. This concept topic examines High Availability (HA) cluster topologies, persistent message spooling, and Dead-Message Queue (DMQ) processing for poison messages.

High Availability (HA) Architecture

Ensuring business continuity in enterprise integration requires robust high-availability mechanisms and message preservation patterns capable of handling system failures, network partitions, and slow application consumers.

For mission-critical production workloads, Event Broker Services deploy in High Availability (HA) triple-node cluster configurations spanning distinct availability zones. An HA cluster consists of an Primary node, a Backup node, and a Monitoring node. The Active node processes all messaging traffic and continuously synchronizes message spool state and queue indexes to the Mate node. If the Active node experiences a hardware fault or network isolation, the cluster executes an immediate, automatic failover. Client applications automatically reconnect to the Mate node without data loss or manual operator intervention.

Guaranteed Messaging and Spool Management

When applications require guaranteed message delivery (Persistent Messaging), events are written directly to non-volatile disk storage (Message Spool) before an acknowledgement is returned to the publisher. Messages remain safely stored in the spool until every subscribing queue acknowledges processing. If a downstream consumer application crashes or goes offline for maintenance, the broker retains the messages in order, delivering them seamlessly as soon as the consumer reconnects.

Dead Message Queues (DMQ) and Poison Message Handling

To prevent unprocessable messages ("poison messages") from stalling integration flows, queues are configured with Dead Message Queues (DMQ) and Time-to-Live (TTL) policies. If a consumer repeatedly fails to acknowledge a message (for example, due to a database constraint error or payload formatting bug), the broker increments a redelivery counter.

Once the maximum redelivery attempt limit is reached, the broker automatically removes the problematic message from the primary queue and routes it to a dedicated Dead Message Queue. Isolating failed messages in a DMQ prevents processing bottlenecks on the main event stream, allowing operations teams to inspect payload errors, resolve underlying system issues, and replay the messages safely.

Note

Dead Message Queues must be created manually in Broker Manager before they can be assigned to a queue, they are not created automatically.